
%config InlineBackend.figure_format = 'retina'
%matplotlib inline    
import seaborn as sns 
sns.set(font= "Kaiti",style="ticks",font_scale=1.4)
import matplotlib
matplotlib.rcParams['axes.unicode_minus']=False # 解决坐标轴的负号显示问题
## 导入需要的库
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import plotly.express as px
from fbprophet import Prophet


usedatadf = pd.read_excel("附件1：赛题A数据.xlsx")
usedatadf.head()


usedatadf.columns

Index(['时间', '基站编号', '小区编号', '本地小区标识', '上行可用的PRB个数', '下行可用的PRB个数',
       '上行PhysicalResourceBlock被使用的平均个数', '下行PhysicalResourceBlock被使用的平均个数',
       '上行PUSCH的PhysicalResourceBlock被使用的平均个数', '小区内的平均用户数', '小区内的最大用户数',
       'RRC连接建立完成次数', 'RRC连接请求次数（不包括重发）', 'ERAB建立成功总次数', 'ERAB建立尝试总次数',
       'ERAB异常释放总次数', 'ERAB正常释放总次数', '系统间切换出ERAB正常释放总次数', 'eNodeB内同频切换出成功次数',
       'eNodeB间同频切换出成功次数', 'eNodeB内同频切换出执行次数', 'eNodeB间同频切换出执行次数',
       'eNodeB内异频切换出成功次数', 'eNodeB间异频切换出成功次数', 'eNodeB内异频切换出执行次数',
       'eNodeB间异频切换出执行次数', '小区PDCP层所发送的下行数据的总吞吐量比特',
       '使缓存为空的最后一个TTI所传的下行PDCP吞吐量比', '扣除使下行缓存为空的最后一个TTI之后的数传时',
       '小区PDCP层所接收到的上行数据的总吞吐量比特', '使UE缓存为空的最后一个TTI所传的上行PDCP吞吐量',
       '扣除使UE缓存为空的最后一个TTI之后的上行数传', '平均激活用户数', '最大激活用户数', '空口上报全带宽CQI为0的次数',
       '空口上报全带宽CQI为1的次数', '空口上报全带宽CQI为2的次数', '空口上报全带宽CQI为3的次数',
       '空口上报全带宽CQI为4的次数', '空口上报全带宽CQI为5的次数', '空口上报全带宽CQI为6的次数',
       '空口上报全带宽CQI为7的次数', '空口上报全带宽CQI为8的次数', '空口上报全带宽CQI为9的次数',
       '空口上报全带宽CQI为10的次数', '空口上报全带宽CQI为11的次数', '空口上报全带宽CQI为12的次数',
       '空口上报全带宽CQI为13的次数', '空口上报全带宽CQI为14的次数', '空口上报全带宽CQI为15的次数',
       '用户随机接入时TA值在区间0范围的接入次数', '用户随机接入时TA值在区间1范围的接入次数',
       '用户随机接入时TA值在区间2范围的接入次数', '用户随机接入时TA值在区间3范围的接入次数',
       '用户随机接入时TA值在区间4范围的接入次数', '用户随机接入时TA值在区间5范围的接入次数',
       '用户随机接入时TA值在区间6范围的接入次数', '用户随机接入时TA值在区间7范围的接入次数',
       '用户随机接入时TA值在区间8范围的接入次数', '用户随机接入时TA值在区间9范围的接入次数',
       '用户随机接入时TA值在区间10范围的接入次数', '用户随机接入时TA值在区间11范围的接入次数',
       'MR测量上报RSRP在Index0区间的次数', 'MR测量上报RSRP在Index1区间的次数',
       'MR测量上报RSRP在Index2区间的次数', 'MR测量上报RSRP在Index3区间的次数',
       'MR测量上报RSRP在Index4区间的次数', '小区QCI为1的DRB业务PDCPSDU上行丢弃的总包数包',
       '小区QCI为1的DRB业务PDCPSDU上行期望收到的总包数', '小区QCI为1的DRB业务PDCPSDU下行空口丢弃的总包数',
       '小区QCI为1的DRB业务PDCPSDU下行空口发送的总包数'],
      dtype='object')


## 数据提取,获取区内的平均用户数数据，指标1
usenumdf = usedatadf.iloc[:,0:4]
usenumdf["number"] = usedatadf["小区内的平均用户数"]
usenumdf.head()


## 可视化平均用户数的数据分布
plt.figure(figsize=(10,6))
sns.kdeplot(data=usenumdf, x="number",bw_adjust=.3,lw = 3,
             hue="基站编号",palette = "Set1") 
plt.title("小区内的平均用户数")
plt.show()


## 数据提取,获取小区PDCP 流量，指标2
PDCPdf = usedatadf.iloc[:,0:4]
PDCPdf["PDCP"] = usedatadf["小区PDCP层所发送的下行数据的总吞吐量比特"]+usedatadf["小区PDCP层所接收到的上行数据的总吞吐量比特"]
PDCPdf.head()


## 可视化小区PDCP流量的数据分布
plt.figure(figsize=(10,6))
sns.kdeplot(data=PDCPdf, x="PDCP",bw_adjust=.3,lw = 3,
             hue="基站编号",palette = "Set1") 
plt.title("小区PDCP流量")
plt.show()


## 数据提取,获取平均激活用户数，指标3
avgnumdf = usedatadf.iloc[:,0:4]
avgnumdf["avgnumber"] = usedatadf["平均激活用户数"]
avgnumdf.head()


## 可视化小区PDCP流量的数据分布
plt.figure(figsize=(10,6))
sns.kdeplot(data=avgnumdf, x="avgnumber",bw_adjust=.3,lw = 3,
             hue="基站编号",palette = "Set1") 
plt.title("平均激活用户数")
plt.show()


mydf1 = usenumdf.groupby(["基站编号","小区编号"]).count().reset_index()
mydf1 =mydf1.groupby(["基站编号"]).count().reset_index()
mydf1.plot(x = "基站编号",y = "number",kind = "bar",figsize= (10,6),
           title = "每个基站的覆盖小区数量")
plt.show()


## 获取一个包含三个指标的数据
myusealldf = usenumdf
myusealldf["PDCP"] = PDCPdf["PDCP"]
myusealldf["avgnumber"] = avgnumdf["avgnumber"]
myusealldf.head()


## 可视化三个指标的相关性
index = np.random.permutation(len(myusealldf))[0:2000] # 可视化部分数据
p = sns.pairplot(myusealldf.iloc[index,[1,4,5,6]], hue="基站编号",
                 markers = ["o", "s", "D","d","*"],
                 palette = "Set1",
                 height=3,aspect=3/2) 
p.map_lower(sns.kdeplot, levels=10) ## 为下三角区域添加密度曲线
plt.show()

## 三个指标之间有较强的相关性


usenumdf.head()


## 可视化平均用户数数据

plt.figure(figsize=(20,10))
sns.lineplot(data=usenumdf,x = "时间",y = "number",hue = "小区编号", palette="tab10", linewidth=2.5)
plt.legend(loc= (1,0.1),ncol=3)
plt.show()
## 不便于观察，可视化部分数据


## 获取所有小区的编号
xiaoqu = np.unique(usenumdf["小区编号"])
xiaoqu

array([26019001, 26019002, 26019003, 26019004, 26019005, 26019006,
       26019007, 26019008, 26019009, 26019010, 26019011, 26019012,
       26019013, 26019014, 26019015, 26019016, 26019017, 26019018,
       26019019, 26019020, 26019021, 26019022, 26019023, 26019024,
       26019025, 26019026, 26019027, 26019028, 26019029, 26019030,
       26019031, 26019032, 26019033, 26019034, 26019035, 26019036,
       26019037, 26019038, 26019039, 26019040, 26019041, 26019042,
       26019043, 26019044, 26019045, 26019046, 26019047, 26019048,
       26019049, 26019050, 26019051, 26019052, 26019053, 26019054,
       26019055, 26019056, 26019057, 26019058])


## 只可视化其中几个小区的数据
[True for element in xiaoqu[0:5] if element in xiaoqu]

[True, True, True, True, True]


plt.figure(figsize=(15,8))
sns.lineplot(data=usenumdf.loc[usenumdf["小区编号"] == 26019001,:],x = "时间",y = "number",
             hue = "小区编号", palette="tab10", linewidth=2.5)
plt.show()


## 获取可交互的图像
fig = px.line(usenumdf.loc[usenumdf["小区编号"] == 26019001,:],x = "时间",y = "number",width=1000, height=600,
              title = "平均用户数")
fig.update_traces(line=dict(width=2, color="black"))
fig.update_layout(title={"x":0.5,"y":0.9})
fig.show()


## 可视化多个小区的数据，获取可交互的图像
fig = px.line(usenumdf.loc[usenumdf["小区编号"] <= 26019008,:],x = "时间",y = "number",width=1000, height=600,
              title = "平均用户数",color = "小区编号")
fig.update_traces(line=dict(width=2))
fig.update_layout(title={"x":0.5,"y":0.9})
fig.show()


## 可视化多个小区的数据，获取可交互的图像
fig = px.line(myusealldf.loc[myusealldf["小区编号"] <= 26019005,:],x = "时间",y = "PDCP",width=1000, height=600,
              title = "PDCP",color = "小区编号")
fig.update_traces(line=dict(width=2))
fig.update_layout(title={"x":0.5,"y":0.9})
fig.show()


## 可视化多个小区的数据，获取可交互的图像
fig = px.line(myusealldf.loc[myusealldf["小区编号"] <= 26019005,:],x = "时间",y = "avgnumber",width=1000, height=600,
              title = "PDCP",color = "小区编号")
fig.update_traces(line=dict(width=2))
fig.update_layout(title={"x":0.5,"y":0.9})
fig.show()


## 数据准备
predf = myusealldf.loc[myusealldf["小区编号"] == 26019005,["时间","number"]]
predf.columns = ["ds","y"]
predf.ds = pd.to_datetime(predf.ds,format = "%Y-%m-%d %H:%M")
predf


## 可视化时间序列的变化情况
predf.plot(x = "ds",y = "y",style = "b-o",figsize=(16,7))
plt.grid()
plt.title("时间序列数据的波动情况")
plt.show()


## 针对前面的时序数据，可以使用下面的程序获得时序数据拟合模型，对数据变化趋势和波动情况进行拟合，并且在模型的预测结果中包含预测值的上界和下界（默认为置信度95%的上下界）。
## 定义一个函数outlier_detection()，该函数会使用模型预测值的置信区间的上下界，来判断样本是否为异常值，
## 根据模型预测值的置信区间"yhat_lower"和"yhat_upper"判断样本是否为异常值
def outlier_detection(forecast):
    index = np.where((forecast["y"] <= forecast["yhat_lower"])|
                     (forecast["y"] >= forecast["yhat_upper"]),True,False)
    return index


## 对该数据建立一个时间序列模型
np.random.seed(1234)  ## 设置随机数种子
model = Prophet(growth="linear",daily_seasonality = True,
                weekly_seasonality=True,yearly_seasonality=False,
                seasonality_mode = 'multiplicative', 
                interval_width = 0.95,   ## 获取95%的置信区间
                )
model = model.fit(predf)     # 使用数据拟合模型
forecast = model.predict(predf)  # 使用模型对数据进行预测
forecast["y"] = predf["y"].reset_index(drop = True)
forecast["yhat"] = np.where(forecast["yhat"]<0,0,forecast["yhat"])
forecast["yhat_lower"] = np.where(forecast["yhat_lower"]<0,0,forecast["yhat_lower"])
forecast[["ds","y","yhat","yhat_lower","yhat_upper"]].head()
## 如果yhat_lower活着yhat小于0，则修改为0

Initial log joint probability = -32.8427
    Iter      log prob        ||dx||      ||grad||       alpha      alpha0  # evals  Notes 
      99       1355.85    0.00166816       83.4561           1           1      132   
    Iter      log prob        ||dx||      ||grad||       alpha      alpha0  # evals  Notes 
     179       1356.53   0.000162473       77.5984   2.582e-06       0.001      276  LS failed, Hessian reset 
     199       1356.54   1.38495e-05       84.9198      0.6384      0.6384      301   
    Iter      log prob        ||dx||      ||grad||       alpha      alpha0  # evals  Notes 
     218       1356.56   3.34127e-06       54.0073   4.784e-08       0.001      369  LS failed, Hessian reset 
     261       1356.57   2.10543e-05       73.9325   3.224e-07       0.001      455  LS failed, Hessian reset 
     299       1356.57   1.06392e-06       64.9042      0.7121      0.7121      506   
    Iter      log prob        ||dx||      ||grad||       alpha      alpha0  # evals  Notes 
     311       1356.57   6.87284e-08       58.6732      0.5036      0.5036      521   
Optimization terminated normally: 
  Convergence detected: relative gradient magnitude is below tolerance


## 查看异常值预测情况
outlier_index = outlier_detection(forecast)
outlier_df = predf[outlier_index]
print("异常值的数量为:",np.sum(outlier_index))

异常值的数量为: 42


outlier_df


## 可视化异常值的结果
fig, ax = plt.subplots()
## 可视化预测值
forecast.plot(x = "ds",y = "yhat",style = "b-",figsize=(16,7),
              label = "预测值",ax=ax)
## 可视化出置信区间
ax.fill_between(forecast["ds"].values, forecast["yhat_lower"], 
                forecast["yhat_upper"],color='b',alpha=.2,
                label = "95%置信区间")
forecast.plot(kind = "scatter",x = "ds",y = "y",c = "k",
              s = 20,label = "原始数据",ax = ax)
## 可视化出异常值的点
outlier_df.plot(x = "ds",y = "y",style = "rs",ax = ax,
                label = "异常值")
plt.legend(loc = 2)
plt.grid()
plt.title("时间序列异常值检测结果")
plt.show()


## 同一数据，不同置信度的异常值查找
predf = myusealldf.loc[myusealldf["小区编号"] == 26019005,["时间","number"]]
predf.columns = ["ds","y"]
predf.ds = pd.to_datetime(predf.ds,format = "%Y-%m-%d %H:%M")
## 对该数据建立一个时间序列模型
np.random.seed(1234)  ## 设置随机数种子
model = Prophet(growth="linear",daily_seasonality = True,
                weekly_seasonality=True,yearly_seasonality=False,
                seasonality_mode = 'multiplicative', 
                interval_width = 0.90,   ## 获取90%的置信区间
                )
model = model.fit(predf)     # 使用数据拟合模型
forecast = model.predict(predf)  # 使用模型对数据进行预测
forecast["y"] = predf["y"].reset_index(drop = True)
forecast["yhat"] = np.where(forecast["yhat"]<0,0,forecast["yhat"])
forecast["yhat_lower"] = np.where(forecast["yhat_lower"]<0,0,forecast["yhat_lower"])
## 查看异常值预测情况
outlier_index = outlier_detection(forecast)
outlier_df = predf[outlier_index]
print("异常值的数量为:",np.sum(outlier_index))
## 可视化异常值的结果
fig, ax = plt.subplots()
## 可视化预测值
forecast.plot(x = "ds",y = "yhat",style = "b-",figsize=(16,7),
              label = "预测值",ax=ax)
## 可视化出置信区间
ax.fill_between(forecast["ds"].values, forecast["yhat_lower"], 
                forecast["yhat_upper"],color='b',alpha=.2,
                label = "90%置信区间")
forecast.plot(kind = "scatter",x = "ds",y = "y",c = "k",
              s = 20,label = "原始数据",ax = ax)
## 可视化出异常值的点
outlier_df.plot(x = "ds",y = "y",style = "rs",ax = ax,
                label = "异常值")
plt.legend(loc = 2)
plt.grid()
plt.title("时间序列异常值检测结果")
plt.show()

Initial log joint probability = -32.8427
    Iter      log prob        ||dx||      ||grad||       alpha      alpha0  # evals  Notes 
      99       1355.85    0.00166816       83.4561           1           1      132   
    Iter      log prob        ||dx||      ||grad||       alpha      alpha0  # evals  Notes 
     179       1356.53   0.000162473       77.5984   2.582e-06       0.001      276  LS failed, Hessian reset 
     199       1356.54   1.38495e-05       84.9198      0.6384      0.6384      301   
    Iter      log prob        ||dx||      ||grad||       alpha      alpha0  # evals  Notes 
     218       1356.56   3.34127e-06       54.0073   4.784e-08       0.001      369  LS failed, Hessian reset 
     261       1356.57   2.10543e-05       73.9325   3.224e-07       0.001      455  LS failed, Hessian reset 
     299       1356.57   1.06392e-06       64.9042      0.7121      0.7121      506   
    Iter      log prob        ||dx||      ||grad||       alpha      alpha0  # evals  Notes 
     311       1356.57   6.87284e-08       58.6732      0.5036      0.5036      521   
Optimization terminated normally: 
  Convergence detected: relative gradient magnitude is below tolerance
异常值的数量为: 66


## 查看一个其他数据
predf = myusealldf.loc[myusealldf["小区编号"] == 26019010,["时间","number"]]
predf.columns = ["ds","y"]
predf.ds = pd.to_datetime(predf.ds,format = "%Y-%m-%d %H:%M")
## 对该数据建立一个时间序列模型
np.random.seed(1234)  ## 设置随机数种子
model = Prophet(growth="linear",daily_seasonality = True,
                weekly_seasonality=True,yearly_seasonality=False,
                seasonality_mode = 'multiplicative', 
                interval_width = 0.95,   ## 获取95%的置信区间
                )
model = model.fit(predf)     # 使用数据拟合模型
forecast = model.predict(predf)  # 使用模型对数据进行预测
forecast["y"] = predf["y"].reset_index(drop = True)
forecast["yhat"] = np.where(forecast["yhat"]<0,0,forecast["yhat"])
forecast["yhat_lower"] = np.where(forecast["yhat_lower"]<0,0,forecast["yhat_lower"])
## 查看异常值预测情况
outlier_index = outlier_detection(forecast)
outlier_df = predf[outlier_index]
print("异常值的数量为:",np.sum(outlier_index))
## 可视化异常值的结果
fig, ax = plt.subplots()
## 可视化预测值
forecast.plot(x = "ds",y = "yhat",style = "b-",figsize=(16,7),
              label = "预测值",ax=ax)
## 可视化出置信区间
ax.fill_between(forecast["ds"].values, forecast["yhat_lower"], 
                forecast["yhat_upper"],color='b',alpha=.2,
                label = "90%置信区间")
forecast.plot(kind = "scatter",x = "ds",y = "y",c = "k",
              s = 20,label = "原始数据",ax = ax)
## 可视化出异常值的点
outlier_df.plot(x = "ds",y = "y",style = "rs",ax = ax,
                label = "异常值")
plt.legend(loc = 2)
plt.grid()
plt.title("时间序列异常值检测结果")
plt.show()

Initial log joint probability = -10.4101
    Iter      log prob        ||dx||      ||grad||       alpha      alpha0  # evals  Notes 
      99        1128.6   0.000306573       83.7598      0.4665           1      139   
    Iter      log prob        ||dx||      ||grad||       alpha      alpha0  # evals  Notes 
     199       1129.15   4.41284e-05       84.1581      0.1782           1      271   
    Iter      log prob        ||dx||      ||grad||       alpha      alpha0  # evals  Notes 
     226       1129.15   2.55137e-08       76.2364     0.04549      0.4854      312   
Optimization terminated normally: 
  Convergence detected: relative gradient magnitude is below tolerance
异常值的数量为: 28


## 同一数据，不同置信度的异常值查找
predf = myusealldf.loc[myusealldf["小区编号"] == 26019010,["时间","number"]]
predf.columns = ["ds","y"]
predf.ds = pd.to_datetime(predf.ds,format = "%Y-%m-%d %H:%M")
## 对该数据建立一个时间序列模型
np.random.seed(1234)  ## 设置随机数种子
model = Prophet(growth="linear",daily_seasonality = True,
                weekly_seasonality=True,yearly_seasonality=False,
                seasonality_mode = 'multiplicative', 
                interval_width = 0.90,   ## 获取90%的置信区间
                )
model = model.fit(predf)     # 使用数据拟合模型
forecast = model.predict(predf)  # 使用模型对数据进行预测
forecast["y"] = predf["y"].reset_index(drop = True)
forecast["yhat"] = np.where(forecast["yhat"]<0,0,forecast["yhat"])
forecast["yhat_lower"] = np.where(forecast["yhat_lower"]<0,0,forecast["yhat_lower"])
## 查看异常值预测情况
outlier_index = outlier_detection(forecast)
outlier_df = predf[outlier_index]
print("异常值的数量为:",np.sum(outlier_index))
## 可视化异常值的结果
fig, ax = plt.subplots()
## 可视化预测值
forecast.plot(x = "ds",y = "yhat",style = "b-",figsize=(16,7),
              label = "预测值",ax=ax)
## 可视化出置信区间
ax.fill_between(forecast["ds"].values, forecast["yhat_lower"], 
                forecast["yhat_upper"],color='b',alpha=.2,
                label = "90%置信区间")
forecast.plot(kind = "scatter",x = "ds",y = "y",c = "k",
              s = 20,label = "原始数据",ax = ax)
## 可视化出异常值的点
outlier_df.plot(x = "ds",y = "y",style = "rs",ax = ax,
                label = "异常值")
plt.legend(loc = 2)
plt.grid()
plt.title("时间序列异常值检测结果")
plt.show()

Initial log joint probability = -10.4101
    Iter      log prob        ||dx||      ||grad||       alpha      alpha0  # evals  Notes 
      99        1128.6   0.000306573       83.7598      0.4665           1      139   
    Iter      log prob        ||dx||      ||grad||       alpha      alpha0  # evals  Notes 
     199       1129.15   4.41284e-05       84.1581      0.1782           1      271   
    Iter      log prob        ||dx||      ||grad||       alpha      alpha0  # evals  Notes 
     226       1129.15   2.55137e-08       76.2364     0.04549      0.4854      312   
Optimization terminated normally: 
  Convergence detected: relative gradient magnitude is below tolerance
异常值的数量为: 55


## 查看一个其他数据
predf = myusealldf.loc[myusealldf["小区编号"] == 26019010,["时间","number"]]
predf.columns = ["ds","y"]
predf.ds = pd.to_datetime(predf.ds,format = "%Y-%m-%d %H:%M")
## 对该数据建立一个时间序列模型
np.random.seed(1234)  ## 设置随机数种子
model = Prophet(growth="linear",daily_seasonality = True,
                weekly_seasonality=True,yearly_seasonality=False,
                seasonality_mode = 'multiplicative', 
                interval_width = 0.98,   ## 获取98%的置信区间
                )
model = model.fit(predf)     # 使用数据拟合模型
forecast = model.predict(predf)  # 使用模型对数据进行预测
forecast["y"] = predf["y"].reset_index(drop = True)
forecast["yhat"] = np.where(forecast["yhat"]<0,0,forecast["yhat"])
forecast["yhat_lower"] = np.where(forecast["yhat_lower"]<0,0,forecast["yhat_lower"])
## 查看异常值预测情况
outlier_index = outlier_detection(forecast)
outlier_df = predf[outlier_index]
print("异常值的数量为:",np.sum(outlier_index))
## 可视化异常值的结果
fig, ax = plt.subplots()
## 可视化预测值
forecast.plot(x = "ds",y = "yhat",style = "b-",figsize=(16,7),
              label = "预测值",ax=ax)
## 可视化出置信区间
ax.fill_between(forecast["ds"].values, forecast["yhat_lower"], 
                forecast["yhat_upper"],color='b',alpha=.2,
                label = "98%置信区间")
forecast.plot(kind = "scatter",x = "ds",y = "y",c = "k",
              s = 20,label = "原始数据",ax = ax)
## 可视化出异常值的点
outlier_df.plot(x = "ds",y = "y",style = "rs",ax = ax,
                label = "异常值")
plt.legend(loc = 2)
plt.grid()
plt.title("时间序列异常值检测结果")
plt.show()

Initial log joint probability = -10.4101
    Iter      log prob        ||dx||      ||grad||       alpha      alpha0  # evals  Notes 
      99        1128.6   0.000306573       83.7598      0.4665           1      139   
    Iter      log prob        ||dx||      ||grad||       alpha      alpha0  # evals  Notes 
     199       1129.15   4.41284e-05       84.1581      0.1782           1      271   
    Iter      log prob        ||dx||      ||grad||       alpha      alpha0  # evals  Notes 
     226       1129.15   2.55137e-08       76.2364     0.04549      0.4854      312   
Optimization terminated normally: 
  Convergence detected: relative gradient magnitude is below tolerance
异常值的数量为: 19


outlier_df

	时间	基站编号	小区编号	本地小区标识	上行可用的PRB个数	下行可用的PRB个数	上行PhysicalResourceBlock被使用的平均个数	下行PhysicalResourceBlock被使用的平均个数	上行PUSCH的PhysicalResourceBlock被使用的平均个数	小区内的平均用户数	...	MR测量上报RSRP在Index0区间的次数	MR测量上报RSRP在Index1区间的次数	MR测量上报RSRP在Index2区间的次数	MR测量上报RSRP在Index3区间的次数	MR测量上报RSRP在Index4区间的次数	小区QCI为1的DRB业务PDCPSDU上行期望收到的总包数	小区QCI为1的DRB业务PDCPSDU下行空口发送的总包数
0	2021-08-28 00:00	1200071	26019014	0	100	100	6.4553	31.5065	2.1503	22.6341	...	3	16	17	178	9943	2218	2179
1	2021-08-28 00:00	1200071	26019015	1	100	100	8.8281	21.5452	4.2220	19.1064	...	2	4	21	180	8427	1510	1423
2	2021-08-28 00:00	1200071	26019016	2	100	100	5.5024	11.9276	1.8618	15.4729	...	2	4	48	575	6371	2030	1919
3	2021-08-28 00:00	1200071	26019023	3	100	100	3.1113	0.9637	0.0077	0.0509	...	0	0	0	3	31	0	0
4	2021-08-28 00:00	1200071	26019024	4	100	100	2.6004	1.0051	0.1201	0.6355	...	0	1	1	9	285	7526	7780

	时间	基站编号	小区编号	本地小区标识	number
0	2021-08-28 00:00	1200071	26019014	0	22.6341
1	2021-08-28 00:00	1200071	26019015	1	19.1064
2	2021-08-28 00:00	1200071	26019016	2	15.4729
3	2021-08-28 00:00	1200071	26019023	3	0.0509
4	2021-08-28 00:00	1200071	26019024	4	0.6355

	时间	基站编号	小区编号	本地小区标识	PDCP
0	2021-08-28 00:00	1200071	26019014	0	32161243768
1	2021-08-28 00:00	1200071	26019015	1	17015521760
2	2021-08-28 00:00	1200071	26019016	2	8961952016
3	2021-08-28 00:00	1200071	26019023	3	343112
4	2021-08-28 00:00	1200071	26019024	4	39529280

	时间	基站编号	小区编号	本地小区标识	avgnumber
0	2021-08-28 00:00	1200071	26019014	0	2.7893
1	2021-08-28 00:00	1200071	26019015	1	3.9198
2	2021-08-28 00:00	1200071	26019016	2	3.1944
3	2021-08-28 00:00	1200071	26019023	3	0.0004
4	2021-08-28 00:00	1200071	26019024	4	0.0263

	时间	基站编号	小区编号	本地小区标识	number	PDCP	avgnumber
0	2021-08-28 00:00	1200071	26019014	0	22.6341	32161243768	2.7893
1	2021-08-28 00:00	1200071	26019015	1	19.1064	17015521760	3.9198
2	2021-08-28 00:00	1200071	26019016	2	15.4729	8961952016	3.1944
3	2021-08-28 00:00	1200071	26019023	3	0.0509	343112	0.0004
4	2021-08-28 00:00	1200071	26019024	4	0.6355	39529280	0.0263

赛题A：智能运维中的异常检测与趋势预测¶

数据导入与预处理¶

数据可视化¶

对其中一个小区的数据建立时间序列模型，并识别是否存在异常值¶

异常值数量往往是少数的情况，因此使用置信度为95%获得的异常值数量，更合适（个人认为），异常区间可以根据附近是否有异常值的聚集进行判断。¶

而且还可以使用其它的时间序列模型进行异常值数据的发现和识别¶

	ds	y
18	2021-08-28 00:00:00	15.5100
76	2021-08-28 01:00:00	13.5003
134	2021-08-28 02:00:00	9.3439
192	2021-08-28 03:00:00	9.3711
250	2021-08-28 04:00:00	8.6753
...	...	...
40096	2021-09-25 19:00:00	29.3806
40154	2021-09-25 20:00:00	21.1153
40212	2021-09-25 21:00:00	20.7158
40270	2021-09-25 22:00:00	19.1725
40328	2021-09-25 23:00:00	16.9786

	ds	y	yhat	yhat_lower	yhat_upper
0	2021-08-28 00:00:00	15.5100	13.832869	3.915429	23.091424
1	2021-08-28 01:00:00	13.5003	12.229369	3.206737	22.191514
2	2021-08-28 02:00:00	9.3439	10.866993	0.886551	20.462700
3	2021-08-28 03:00:00	9.3711	9.338182	0.000000	19.411900
4	2021-08-28 04:00:00	8.6753	8.043085	0.000000	18.041248

	ds	y
7964	2021-09-02 17:00:00	47.0583
10226	2021-09-04 08:00:00	24.5867
11676	2021-09-05 09:00:00	28.7775
15794	2021-09-08 08:00:00	49.7990
16316	2021-09-08 17:00:00	53.1923
17708	2021-09-09 17:00:00	55.2697
18578	2021-09-10 08:00:00	51.5614
18810	2021-09-10 12:00:00	52.9775
21478	2021-09-12 10:00:00	34.8264
22754	2021-09-13 08:00:00	50.0628
23276	2021-09-13 17:00:00	57.4975
24668	2021-09-14 17:00:00	54.0883
26060	2021-09-15 17:00:00	57.4686
27452	2021-09-16 17:00:00	55.9756
28554	2021-09-17 12:00:00	54.7347
29946	2021-09-18 12:00:00	52.3993
30062	2021-09-18 14:00:00	48.0720
30120	2021-09-18 15:00:00	55.2931
30178	2021-09-18 16:00:00	55.4046
30236	2021-09-18 17:00:00	47.1689
31222	2021-09-19 10:00:00	32.3628
32556	2021-09-20 09:00:00	31.6681
32614	2021-09-20 10:00:00	30.8867
32672	2021-09-20 11:00:00	28.8367
32730	2021-09-20 12:00:00	22.8125
32788	2021-09-20 13:00:00	23.4969
32846	2021-09-20 14:00:00	25.0133
32904	2021-09-20 15:00:00	21.9967
32962	2021-09-20 16:00:00	22.5136
33020	2021-09-20 17:00:00	25.6225
34122	2021-09-21 12:00:00	23.4668
34180	2021-09-21 13:00:00	20.8497
34238	2021-09-21 14:00:00	22.3478
34296	2021-09-21 15:00:00	26.2102
34354	2021-09-21 16:00:00	24.1259
34412	2021-09-21 17:00:00	21.6572
34470	2021-09-21 18:00:00	20.8781
35804	2021-09-22 17:00:00	50.7489
37196	2021-09-23 17:00:00	49.7456
38414	2021-09-24 14:00:00	51.4728
38472	2021-09-24 15:00:00	52.0725
38588	2021-09-24 17:00:00	57.2870

	ds	y
4587	2021-08-31 07:00:00	8.6683
5573	2021-09-01 00:00:00	7.6025
17173	2021-09-09 08:00:00	9.0672
20479	2021-09-11 17:00:00	7.4256
22741	2021-09-13 08:00:00	10.6269
24133	2021-09-14 08:00:00	12.7124
25467	2021-09-15 07:00:00	9.6444
26917	2021-09-16 08:00:00	1.8006
27091	2021-09-16 11:00:00	9.2128
27323	2021-09-16 15:00:00	9.9558
27381	2021-09-16 16:00:00	12.7292
28309	2021-09-17 08:00:00	1.4796
28483	2021-09-17 11:00:00	10.3278
28715	2021-09-17 15:00:00	11.0772
28773	2021-09-17 16:00:00	13.5164
29701	2021-09-18 08:00:00	10.7010
34109	2021-09-21 12:00:00	8.3136
36603	2021-09-23 07:00:00	10.2212
40025	2021-09-25 18:00:00	7.7242